A machine learning model for the prediction of survival and tumor subtype in pancreatic ductal adenocarcinoma from preoperative diffusion-weighted imaging

European Radiology Experimental

Table 1 Acquisition parameters for the training and independent validation cohorts

	Training cohort	Independent validation cohort
System	Siemens Magnetom Avanto	Siemens Biograph mMR
Software version	VB17	VB18
Anatomic sequences	Axial and coronal T2-weighted HASTE, 5-mm thickness	Axial and coronal T2-weighted HASTE, 5-mm thickness
	Axial T1-weighted VIBE, 3-mm thickness	Axial T1-weighted VIBE, 3-mm thickness
Dynamic study	Axial T1-weighted SPAIR	Axial T1-weighted T1 SPAIR
DWI acquisition	Axial low-resolution EPI, b = 0, 50, 300, 600 s/mm²	Axial low-resolution EPI, b = 0, 50, 300, 600 s/mm²
ADC fit	Linear, b = 50, 300, 600	Linear b = 50, 300, 600
Acquisition voxel size (x, y, z)	5.5 × 5.5 × 5.5 mm	5.1 × 5.1 × 5.1 mm
ADC reconstruction matrix	192 × 192	192 × 192
ADC field of view	360 × 360	360 × 360

ADC Apparent diffusion coefficient, DWI Diffusion-weighted imaging, HASTE Half-Fourier acquisition single-shot turbo spin-echo, SPAIR Spectral attenuated inversion recovery, VIBE Volume interpolated breath-hold examination